Coastal Dynamics 2017 Paper No. 137 1452 FLASH RIP STATISTICS FROM VIDEO IMAGES Guy Rodier Mabiala 1 , France Floc’h 2 , Rafaël Almar 3 , Bruno Castelle 4 , Nicholas Halls 3 , Yves Du Penhoat 13 and Tim Scott 5 Abstract The coastal area of the Gulf of Guinea is vulnerable to the phenomenon of coastal erosion. Many authors already proved that rip currents play a crucial role in coastal morphodynamic and erosion processes. The coexistence of Swash and Flash rip-currents at Grand-Popo beach was shown using lagrangian buoys measurements. In order to study the hydrodynamic conditions related to the generation of flash rip currents, a method based on color processing of the video image has been set up to detect the presence of rip-currents via a change of turbidity, to extract their characteristics and to study their link with forcings. During our study, 434 events of the rip-currents were counted over one short period of 7 days. The majority of rips occurs at low tide as already mentioned in literature, migrates down- drift. Flash rip activity was maximized for shore-normal wave incidence and significant wave height of 1.2-1.5 m. Key words: hydrodynamics; Flash-rip; video imaging; Grand-Popo beach; erosion; coastal. 1. Introduction RIP currents are cross-shore coastal currents. They are narrow and oriented offshore with high speeds (about a m/s) and often appear along steep sandy beaches dominated by waves (or swells). According to the review of Castelle et al. (2016), they are now classified in three main types: focused, fixed, and flash rips. Focused and fixed rips are coupled with three -dimensional bathymetric features and forced by hard structures in the surf zone, respectively, while flash rips are generated by transient nearshore flow instabilities. The focused and fixed rips are quite well described in the literature while flash rips are poorly documented. The majority of authors converge towards two principal mechanisms based respectively on shearing instabilities of the longshore current and the horizontal vorticity induced by the wave breaking spatial variations. The forcing terms in linear wave theory are defined as the radiation stress by Longuet- Higgins and Stewart (1964). Rip-currents are an important mechanism for the cross-shore exchange of water, nutrient, larvae, pollutant and sediment between the surf zone and the upper shelf (Shanks et al, 2010). These currents also contribute to sediment transport (Cooke, 1970; Komar, 1971; Shorts, 1999) often in great quantity (Brander R.W, 1999; Inman D. and al., 1971). They thus play a crucial role in coastal morphodynamic and erosion processes (Shorts, 1992) especially during storms (Thornton and al., 2007; Birrien and al., 2013). These currents represent the principal death hasard by drowning, for beach users around the world. It has been reported that the African area has the highest drowning rate in the world (Peden and McGee, 2003; WHO, 2010). However the occurrence and the type of rip developing on African beaches remains poorly documented. Using Lagrangian drifters released in the surf zone, Castelle and al. (2013) has shown the coexistence of swash and flash rips on the beach of Grand-popo. Since the flash rips are transitory in time and space, field data on their dynamics is sparse, and video-imaging appears to be a suitable approach to capture their dynamics on large temporal and space scales. But the few direct observations of this kind of current, on the Gold Coast in Australia, revealed a low probability of capture (0, 52%) over a single period of observation (Murray and al., 2013). Since most of world littorals, The coastal area of the Gulf of Guinea 1 CIPMA & IRHOB, Cotonou, Bénin - [email protected]; yves.du-[email protected]2 Domaines Océaniques UMR6538/UBO/IUEM, 29280 Plouzané, France – france.floch@univ-brest.fr 3 IRD-LEGOS (CNRS/IRD/CNES/UPS) 31400 Toulouse, France – [email protected]; [email protected]-mip.fr 4 CNRS/Unv. Bordeaux, UMR EPOC, Allée Geoffroy Saint-Hilaire, France - bruno.castelle @ u-bordeaux.fr 5 University of Plymouth, Drake Circus, Plymouth, PL4 8AA, UK - [email protected]
10
Embed
Coastal Dynamics 2017coastaldynamics2017.dk/onewebmedia/137_MABIALA_GuyRodier.pdf · Guy Rodier Mabiala1, France Floc’h 2, Rafaël Almar 3, Bruno Castelle4, Nicholas Halls3, Yves
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Coastal Dynamics 2017
Paper No. 137
1452
FLASH RIP STATISTICS FROM VIDEO IMAGES
Guy Rodier Mabiala1, France Floc’h
2, Rafaël Almar
3, Bruno Castelle
4, Nicholas Halls
3, Yves Du Penhoat
13
and Tim Scott5
Abstract
The coastal area of the Gulf of Guinea is vulnerable to the phenomenon of coastal erosion. Many authors already
proved that rip currents play a crucial role in coastal morphodynamic and erosion processes. The coexistence of Swash
and Flash rip-currents at Grand-Popo beach was shown using lagrangian buoys measurements. In order to study the
hydrodynamic conditions related to the generation of flash rip currents, a method based on color processing of the
video image has been set up to detect the presence of rip-currents via a change of turbidity, to extract their
characteristics and to study their link with forcings. During our study, 434 events of the rip-currents were counted over
one short period of 7 days. The majority of rips occurs at low tide as already mentioned in literature, migrates down-
drift. Flash rip activity was maximized for shore-normal wave incidence and significant wave height of 1.2-1.5 m.
Key words: hydrodynamics; Flash-rip; video imaging; Grand-Popo beach; erosion; coastal.
1. Introduction
RIP currents are cross-shore coastal currents. They are narrow and oriented offshore with high speeds
(about a m/s) and often appear along steep sandy beaches dominated by waves (or swells). According to
the review of Castelle et al. (2016), they are now classified in three main types: focused, fixed, and flash
rips. Focused and fixed rips are coupled with three-dimensional bathymetric features and forced by hard
structures in the surf zone, respectively, while flash rips are generated by transient nearshore flow
instabilities. The focused and fixed rips are quite well described in the literature while flash rips are poorly
documented. The majority of authors converge towards two principal mechanisms based respectively on
shearing instabilities of the longshore current and the horizontal vorticity induced by the wave breaking
spatial variations. The forcing terms in linear wave theory are defined as the radiation stress by Longuet-
Higgins and Stewart (1964).
Rip-currents are an important mechanism for the cross-shore exchange of water, nutrient, larvae, pollutant
and sediment between the surf zone and the upper shelf (Shanks et al, 2010). These currents also contribute
to sediment transport (Cooke, 1970; Komar, 1971; Shorts, 1999) often in great quantity (Brander R.W,
1999; Inman D. and al., 1971). They thus play a crucial role in coastal morphodynamic and erosion
processes (Shorts, 1992) especially during storms (Thornton and al., 2007; Birrien and al., 2013). These
currents represent the principal death hasard by drowning, for beach users around the world. It has been
reported that the African area has the highest drowning rate in the world (Peden and McGee, 2003; WHO,
2010). However the occurrence and the type of rip developing on African beaches remains poorly
documented. Using Lagrangian drifters released in the surf zone, Castelle and al. (2013) has shown the
coexistence of swash and flash rips on the beach of Grand-popo. Since the flash rips are transitory in time
and space, field data on their dynamics is sparse, and video-imaging appears to be a suitable approach to
capture their dynamics on large temporal and space scales. But the few direct observations of this kind of
current, on the Gold Coast in Australia, revealed a low probability of capture (0, 52%) over a single period
of observation (Murray and al., 2013). Since most of world littorals, The coastal area of the Gulf of Guinea
1 CIPMA & IRHOB, Cotonou, Bénin - [email protected] ; [email protected] 2 Domaines Océaniques UMR6538/UBO/IUEM, 29280 Plouzané, France – [email protected] 3 IRD-LEGOS (CNRS/IRD/CNES/UPS) 31400 Toulouse, France – [email protected] ; [email protected] 4CNRS/Unv. Bordeaux, UMR EPOC, Allée Geoffroy Saint-Hilaire, France - bruno.castelle @ u-bordeaux.fr 5 University of Plymouth, Drake Circus, Plymouth, PL4 8AA, UK - [email protected]
is vulnerable to the phenomenon of coastal erosion. With a view to understand the hydrodynamic process
related to the generation of flash rip currents, we set up a method based on image processing in order to
detect the presence of rip-currents according to the surface turbidity, to extract their characteristics and to
study their link with hydrodynamic forcings. The general objective is to analyze the hydrodynamic
conditions related to the generation of flash rip currents.
2. METHODS
2.1. Study site
Grand Popo is an open wave-dominated and microtidal beach (mean spring tide range:~1.8 m) exposed to long period swells with a mean significant wave height 𝐻𝑠 = 1.36𝑚 and a mean peak wave period
𝑇𝑃 = 9.4𝑠 . The combination of the medium to coarse quartz sand ( 0.4– 1𝑚𝑚 , 𝐷50: 0.6 𝑚𝑚 ) and
dominant groundswell regime generated in the South Atlantic results in a modal intermediate, somewhat
reflective, beach state corresponding to the low-tide terrace state following Wright & Short (1984). The
combined effect of persistent swells throughout the year and beach steepness results in an intense easterly
longshore drift of about 0.8 × 106 𝑚3/𝑦𝑒𝑎𝑟 (Laïbi et al., 2013). During the 10-day field campaign, the
wave and tide conditions are obtained from an ADCP moored in 12m of depth, beyond the surf zone. Video
footage were acquired at 2𝐻𝑧, from a video system mounted on a 15 m-high tower, located at about 80𝑚
from the mean waterline.
Figure 1: (a) Location of the Grand Popo beach; (b) Vertical profile of beach; (c) field of study; (d) tower and video
system.
2.2. Experimental Methods
2.2.1. Images rectification
A orthocorrection was applied to the images in order to see a projection of the image in a horizontal plan.
This correction consists on:
cutting out (Figure 2.a) on each instantaneous image the zone of interest of better pixel resolution
following the longshore direction and including the surf and swash zone;
correcting the image distortion due to the lens;
georeferencing the images, according to the ground control point of Grand-Popo taken with a
DGPS-RTK (Origin: foot of the semaphore).
The final image obtained presents a plan view of the site and informs us directly about the real positions
compared to the ground control point of the study (Figure 2 b).
(a)
(b)
(c)
(d)
Coastal Dynamics 2017
Paper No. 137
1454
Figure 2: (a) Interest zone determination ; (b) Rectified and georeferenced image.
Over seven days of video measurement (March 12th-18th, 2014), after quality control, considering only the
clearest daylight hours (from 8:00 to 16:00), 57 hours of images has been processed.
2.2.2. Image processing
The detection of rip-currents on the rectified instantaneous images is a very long task if done manually
image by image. This is due to the quantity of images (57 × 60 × 120 = 410400 images) to visualize. Thus
an automatic method has been developed and set up, based on color processing of images. As the flash rip-
currents transport a large amount of sediments in suspension offshore beyond the surf zone, the surface
water is browner within the rip than in the surrounding sea (which is greener). Detection of rip-currents is
thus based on treatment of the colors of images: initially a threshold on the red in RGB format is used to
remove the part of beach and swash zone in the image. In order to reinforce the contrast between green and
brown outside the surf zone, a threshold on color in HSV-format is applied (Figure 3.c).
The automatic detection of flash rip-currents is carried out on a longshore time-stack (Figure 3.b), over 10
minutes in format HSV, starting from the rectified images. In order to detect the flash rip at different
positions during its lifetme, four stacks were positioned in parallel (Figure 3a). Once detected, a visual
control was done to confirm the presence of a flash rip current. Then morphology characteristics of the
flash rip were computed from a set of instantaneous images (from the beginning to the end of detection, the
complete estimated lifetime of the flash-rip).
Figure 3: (a) Initial image with long-shore stacks positions; (b) Example of long-shore time-stacks showing rip current
(c) Processed image with five characteristics points.
(b)
(a)
(b)
(c)
(a)
Coastal Dynamics 2017
Paper No. 137
1455
The computed characteristics are their positions (cross-shore and longshore position of the feet and the
head, at the beginning and at the end), their extensions (cross-shore and longshore extension) and their
migrations longshore in time, according to the detection of the 5 points (A, B, C, D and E) illustrated on
figure 3 (c).
𝑨 corresponds to the waterline in front of the detected flash-rip. 𝑩 is the first point where the turbidity
flume is detected, on the breaking line. 𝑪 is the most offshore point of the rip head and 𝑬 and 𝑫 define the
longshore extension of the rip head. Then the beginning position (cross-shore ; longshore) is: (𝑿𝑨; 𝒀𝑨);
the cross-shore extension is 𝑿𝑪 − 𝑿𝑨 and the longshore extension is 𝒀𝑬 − 𝒀𝑫.
The precision of the position corresponds to the resolution in real coordinates (0.5 𝑚). The precision of the
lifetime corresponds to the capture frequency (0.5 s). However, the uncertainties might be higher because
of the uncertainties in the visual detection of the brown shape of the rip, not validated by current
measurement, and thus difficult to estimate. Detection is sometimes impossible because of sunshine on
some images leading to a white sea surface.
The mean migration velocity is calculated from the longshore migration of the head and the lifetime. The
occurrence of flash rip during daylight is then correlated with the forcing parameters: waves and tide.
2.2.3. Hydrodynamic parameters
The directional waves spectra as well as the associated averaged parameters ( significant wave height Hs,
peak period Tp, peak direction Dir) were calculated from the orbital velocities measured by the four beams
of an ADCP located 800 m offshore the beach, on 20 minutes burst of acquisition each hour. Hs
corresponds to the significant height of the waves defined as the average of the distances peak-hollow of
the third of the highest waves, the significant height of the component swells (Hs_swell) is calculated from
the spectrum for frequencies going from 0.04 Hz to 0.1 Hz and the significant height of the component
wind sea (Hs_wind), for frequencies higher than 0.1 Hz. Direction is represented in the following by the
incidence according to the cross-shore normal of the beach. The angle between the true North and the
cross-shore direction (oriented offshore) is 172°. Thus waves with normal incidence (incidence of 0°)
corresponds to a wave direction of 172°. SW waves have positive incidence and SE waves have a negative
incidence.
The directional spreading of wave energy is defined as the standard deviation, in radians, of the spectral
width in the limit of a narrow spectrum (Kuik et al., 1988):
With (1.2)
(1.3)
Where 𝜽 = wave direction ; 𝒇 = wave frequency ; 𝑬 = Wave spectral energy density (frequency-
directional wave spectrum) ; 𝒂𝟏𝑒𝑡 𝒃𝟏 = terms of low-order Fourier moments of the frequency-directional
wave spectrum. The mean wave direction at frequency 𝒇 is,
𝜽𝒎(𝒇) = 𝒂𝒓𝒄𝒕𝒂𝒏(𝒂𝟏(𝒇) 𝒃𝟏(𝒇)⁄ ) . (1.4)
𝒂𝟏(𝒇) =∫ 𝐜𝐨𝐬 𝜽𝑬(𝒇,𝜽) 𝒅𝜽
𝟐𝝅𝟎
∫ 𝑬(𝒇,𝜽)𝟐𝝅
𝟎 𝒅𝜽 ;
𝒃𝟏(𝒇) =∫ 𝐜𝐨𝐬 𝜽𝑬(𝒇,𝜽) 𝒅𝜽
𝟐𝝅𝟎
∫ 𝑬(𝒇,𝜽)𝟐𝝅
𝟎 𝒅𝜽 ;
𝝈𝜽(𝒇) = 𝟐 [𝟐 (𝟏 − √𝒂𝟏𝟐(𝒇) + 𝒃𝟏
𝟐(𝒇))]
𝟏 𝟐⁄
(1.1)
Coastal Dynamics 2017
Paper No. 137
1456
3. Results and Discussion
The waves and tide conditions, during the observations leading to our results, were very varied. Shifting
from a neap tide (0.45-m range) to a spring tide (1.2-m range) cycle, Grand Popo beach was exposed to a
waves regime of average incidence SW with an average significant wave height of 1.42 m and an average